Find Active customers in past X days - sql

Question

I am facing some hard times, Need quick help. It would be great if someone could assist me.
Thanks a lot in advance:)
I have 2 tables.
1st table: daily_customer_snapshot: the daily snapshot of the customer which looks something as shown below.
c_id
date
state
location
b1
2020-12-01
Active
OOW
b1
2020-12-02
Active
OOW
b1
2020-12-03
Active
OOW
b1
2020-12-04
Active
OOW
b1
2020-12-05
Active
OOW
b3
2020-12-06
Active
OOW
b3
2020-12-07
Active
OOW
b3
2020-12-08
Active
OOW
b1
2020-12-09
Decay
IW
b2
2020-12-15
Active
OOW
2nd table: customer_date_series: contains the date series from the day user became our customer.
Ex: refer image 2: user b1 became our customer on '2020-12-01' and user b3 became our customer on '2020-12-06'
and b2 became our customer on '2020-12-15'. I have generated the date series with customer_id to count at any given date how many customers we had.
c_id
date
b1
2020-12-01
b1
2020-12-02
b1
2020-12-03
b1
2020-12-04
b1
2020-12-05
b1
2020-12-06
b1
2020-12-07
b1
2020-12-08
b1
2020-12-09
b1
2020-12-10
b1
2020-12-11
b1
2020-12-12
b1
2020-12-13
b1
2020-12-14
b1
2020-12-15
b1
2020-12-16
b3
2020-12-06
b3
2020-12-07
b3
2020-12-08
b3
2020-12-09
b3
2020-12-10
b3
2020-12-11
b3
2020-12-12
b3
2020-12-13
b3
2020-12-14
b3
2020-12-15
b3
2020-12-16
b2
2020-12-15
b2
2020-12-16
I left joined table1 (customer_date_series) with table2 (daily_customer_snapshot) to get the overview of the customer behavior at any given date.
I got the results as displayed in image 3.
Query to Join:
select
bds.date,
bds.c_id,
b.state,
b.location
FROM
customer_date_series bds LEFT JOIN daily_customer_snapshot b ON bds.c_id = b.c_id and bds.date = b.date
ORDER BY
1,2;
date
c_id
state
location
2020-12-01
b1
Active
OOW
2020-12-02
b1
Active
OOW
2020-12-03
b1
Active
OOW
2020-12-04
b1
Active
OOW
2020-12-05
b1
Active
OOW
2020-12-06
b1
2020-12-06
b3
Active
OOW
2020-12-07
b1
2020-12-07
b3
Active
OOW
2020-12-08
b1
2020-12-08
b3
Active
OOW
2020-12-09
b1
Decay
IW
2020-12-09
b3
2020-12-10
b1
2020-12-10
b3
2020-12-11
b1
2020-12-11
b3
2020-12-12
b1
2020-12-12
b3
2020-12-13
b1
2020-12-13
b3
2020-12-14
b1
2020-12-14
b3
2020-12-15
b1
2020-12-15
b2
Active
OOW
2020-12-15
b3
2020-12-16
b1
2020-12-16
b2
2020-12-16
b3
This is where I am struggling.
I am facing a challenge here. I want to create new column called 'status' and if the customer data in the daily_customer_snapshot is updated in the past 5 days from the current_date
I want to set the status to be 'Active' Else 'Inactive'.
Ex:

Answer 1

If I follow you correctly, you can use boolean window aggregation:
select
bds.date,
bds.c_id,
b.state,
b.location,
bool_or(b.state = 'Active') over(
partition by bds.c_id
order by bds.date
range between interval '5 days' preceding and current row
) as is_active
from customer_date_series bds
left join daily_customer_snapshot b on bds.c_id = b.c_id and bds.date = b.date
order by 1,2;
This sets a boolean flag on rows where the same customer was active at least once within the last 5 days (or in the current day).
If you do want to see 'Active'/ 'InActive' instead (which I find less useful than a boolean) you can do:
min(b.state) over(
partition by bds.c_id
order by bds.date
range between interval '5 days' preceding and current row
) as status
... Which works because, string-wise, 'Active' < 'InActive'.

Answer 2

If you want to use both tables, then a lateral join does what you want:
select bds.date, bds.c_id, b.state, b.location
--CASE WHEN b.state = '%ActiveDecay%' between current_date- 10 and current_date THEN 'ActIve' ELSE 'DECAY' END as STATUS
FROM battery_date_series bds LEFT JOIN LATERAL
(SELECT b.*
FROM battery b
WHERE bds.c_id = b.c_id and b.date <= bds.date
ORDER BY b.date DESC
LIMIT 1
) b
ON 1=1
ORDER BY 1,2;